Robust biodegradable synapse with sub-biological energy and extended memory for intelligent reflexive system

Yoojin Chang, Sangyun Na, Yun Goo Ro, Cheolhong Park, Seokhee Jung, Yong‐Jin Park, Min Sub Kwak, Jeeyoon Kim, Hye Jeong Oh, Jaeho Kim, Hyunhyub KoNature Communications 2025 · DOI

Abstract

Biodegradable artificial synapses hold great promise for sustainable neuromorphic electronics, yet combining long-term memory, ultralow energy consumption, and mechanical robustness remains challenging. Here, we report a fully biodegradable multilayer artificial synapse (M-AS) composed of crosslinked chitosan-guar gum (CS-GG) ion-active layers (IALs) and a cellulose acetate (CA) ion-binding layer (IBL). This trilayer architecture enhances ion trapping via ion-dipole coupling (IDC) at the IAL-IBL interface, while hydrogen-bonded crosslinking within the CS-GG matrix enhances mechanical and environmental stability. Sodium chloride, embedded in the IALs, serves as a mobile ionic species analogous to biological neurotransmitters, enabling low-voltage ion migration. Upon electrical stimulation, ion migration and dipole alignment induce IDC, leading to partial ion retention and cascade-like postsynaptic current responses that support memory formation. The M-AS supports key synaptic functionalities—including paired-pulse facilitation, short-term and long-term plasticity, multilevel memory encoding, and bidirectional modulation—under sub-millivolt operation. It achieves the longest long-term memory time (5944 s) reported among biodegradable artificial synapses and an energy consumption (0.85 fJ/event) lower than that of biological synapses. Integration with a thermistor and robotic actuator enables a bioinspired reflexive system capable of adaptive, stimulus-dependent learning and reflex-like behaviors.

Addressing the Sustainability Crisis in Neuromorphic Engineering

Neuromorphic engineering aims to mimic the brain's efficiency, but the field faces a sustainability crisis. Most high-performance artificial synapses rely on non-biodegradable metals (like gold or silver) and rigid substrates that contribute to electronic waste. While biodegradable alternatives exist, they often suffer from a "triple threat" of weaknesses: high energy consumption, physical fragility (especially in humid environments), and fleeting memory retention. This research introduces a synapse that is not only fully compostable but also outperforms previous biodegradable benchmarks in energy efficiency and stability.

Methods: The Multilayer Artificial Synapse (M-AS)

The researchers developed a vertically stacked trilayer configuration designed to optimize ion dynamics:

  1. Ion-Active Layers (IALs): Composed of a crosslinked blend of chitosan (derived from crustacean shells) and guar gum (from beans). These layers are infused with sodium chloride (NaCl), where sodium ions act as mobile "neurotransmitters."
  2. Ion-Binding Layer (IBL): A thin middle layer of cellulose acetate that acts as a "receptor."

The functional breakthrough lies in Ion-Dipole Coupling (IDC). When a voltage is applied, ions migrate and become trapped at the interface between the layers due to electrostatic interactions with the polar groups in the cellulose acetate. This trapping mechanism is significantly more stable than the transient "electric double layers" used in previous devices. To ensure mechanical and environmental robustness, the team utilized hydrogen-bonded crosslinking to create a double-network polymer matrix, preventing the device from dissolving or degrading prematurely when exposed to moisture or heat.

The concept and working mechanism of the M-AS compared to a biological synapse, illustrating the trilayer structure and ion-dipole coupling.

Key Insights and Hardware Characterization

Unlike software-based neuromorphic studies, this hardware-focused research relied on electrochemical characterization to evaluate performance:

  • Sub-Biological Energy Consumption: The synapse operates at microvolt levels, consuming only 0.85 fJ per event. For context, a biological synapse typically uses 1–10 fJ. This efficiency is achieved through highly mobile Na+ ions and a design that minimizes leakage current.
  • Memory Scaling: The researchers discovered that memory retention scales with the number of layers. By stacking 10 IAL-IBL pairs, they extended the memory time to 5944 seconds—the longest ever reported for a biodegradable synapse.
  • Reflexive Logic Integration: The M-AS allows for adaptive learning. In a robotic hand integration, a mild heat stimulus might trigger only a short-term response, but repeated or extreme heat causes the synapse to transition from short-term to long-term memory, triggering a permanent "withdrawal" reflex.

Performance Results

  • Energy Efficiency: 0.85 fJ per synaptic event, outperforming biological synapses.
  • Memory Retention: 5944 seconds (nearly 100 minutes) in a 10-layer configuration.
  • Biodegradability: The entire device, including magnesium electrodes, completely decomposed in natural soil within 16 days without leaving toxic byproducts.
  • Functional Reflexes: Integrated into a robotic hand, the M-AS successfully distinguished between mild heat (requiring no action) and dangerous temperatures (triggering an immediate release).
Comparison of different structural approaches showing how the IAL-IBL-IAL and stacked structures lead to lower energy and significantly higher LTM time.

Limitations and Future Outlook

While the M-AS represents a significant advancement, several challenges remain for large-scale implementation. Scaling a 10-layer stack to the millions of synapses required for a full "brain-on-a-chip" presents manufacturing hurdles. Additionally, the ionic migration mechanism is inherently slower than electron-based switching in traditional silicon, making it ideal for human-scale reflexive systems but potentially unsuitable for high-speed computing. Future work will focus on balancing durability with rapid biodegradability for specific environmental applications.

Summary updated

Abstract

Biodegradable artificial synapses hold great promise for sustainable neuromorphic electronics, yet combining long-term memory, ultralow energy consumption, and mechanical robustness remains challenging. Here, we report a fully biodegradable multilayer artificial synapse (M-AS) composed of crosslinked chitosan-guar gum (CS-GG) ion-active layers (IALs) and a cellulose acetate (CA) ion-binding layer (IBL). This trilayer architecture enhances ion trapping via ion-dipole coupling (IDC) at the IAL-IBL interface, while hydrogen-bonded crosslinking within the CS-GG matrix enhances mechanical and environmental stability. Sodium chloride, embedded in the IALs, serves as a mobile ionic species analogous to biological neurotransmitters, enabling low-voltage ion migration. Upon electrical stimulation, ion migration and dipole alignment induce IDC, leading to partial ion retention and cascade-like postsynaptic current responses that support memory formation. The M-AS supports key synaptic functionalities-including paired-pulse facilitation, short-term and long-term plasticity, multilevel memory encoding, and bidirectional modulation-under sub-millivolt operation. It achieves the longest long-term memory time (5944 s) reported among biodegradable artificial synapses and an energy consumption (0.85 fJ/event) lower than that of biological synapses. Integration with a thermistor and robotic actuator enables a bioinspired reflexive system capable of adaptive, stimulus-dependent learning and reflex-like behaviors. These results demonstrate the potential of M-AS for low-power, intelligent human-machine interfaces.
BibTeX
@inproceedings{chang2025robust,
  author = {Yoojin Chang and Sangyun Na and Yun Goo Ro and Cheolhong Park and Seokhee Jung and Yong‐Jin Park and Min Sub Kwak and Jeeyoon Kim and Hye Jeong Oh and Jaeho Kim and Hyunhyub Ko},
  title = {{Robust biodegradable synapse with sub-biological energy and extended memory for intelligent reflexive system}},
  booktitle = {Nature Communications},
  year = {2025},
  doi = {10.1038/s41467-025-66511-3},
}
Figure